mistralrs bench
Run performance benchmarks for base or LoRA model generation
mistralrs bench [OPTIONS] [COMMAND]| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | HuggingFace model ID or local path to model directory | |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--format <FORMAT> | Model format: plain (safetensors), gguf, or ggml Auto-detected if not specified Possible values: plain, gguf, ggml. | |
-f, --quantized-file <QUANTIZED_FILE> | Quantized model filename(s) for GGUF/GGML (semicolon-separated for multiple) | |
--tok-model-id <TOK_MODEL_ID> | Model ID for tokenizer when using quantized format | |
--gqa <GQA> | 1 | GQA value for GGML models |
--enable-lora | false | Enable dynamic LoRA without preloading an adapter. Supports text models. Qwen3.5/3.6 MoE requires automatic model selection; vision-tower adapters are unsupported |
--lora <ALIAS=SOURCE|JSON> | Preload a language-model LoRA adapter as ALIAS=SOURCE. Remote adapters use revision main. May be repeated. Qwen3.5/3.6 MoE conditional-generation models require auto model selection; vision-tower adapters are unsupported | |
--lora-max-adapters <LORA_MAX_ADAPTERS> | 16 | Maximum loaded LoRA aliases and, independently, resident adapter generations |
--lora-max-rank <LORA_MAX_RANK> | 256 | Maximum rank accepted for a LoRA adapter |
--lora-max-bytes <BYTES> | 8589934592 | Maximum memory used by loaded adapters |
--legacy-lora <SOURCE> | Legacy LoRA adapter source for a raw GGUF or GGML model | |
--legacy-lora-order <LEGACY_LORA_ORDER> | Ordering JSON file for a legacy raw GGUF or GGML LoRA adapter | |
--xlora <XLORA> | X-LoRA adapter model ID | |
--xlora-order <XLORA_ORDER> | X-LoRA ordering JSON file | |
--tgt-non-granular-index <TGT_NON_GRANULAR_INDEX> | Target non-granular index for X-LoRA | |
--quant <QUANT> | Quantization front-door: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) This prefers prebuilt UQFF from mistralrs-community/<model>-UQFF, so use --isq if you do not want to switch to a prebuilt UQFF | |
--isq <IN_SITU_QUANT> | In-situ quantization: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) and quantizes the selected model in-place (in-situ) | |
--from-uqff <FROM_UQFF> | UQFF file(s) to load from. Accepts numeric shorthands (2, 3, 4, 5, 6, 8) to auto-detect the appropriate UQFF file (e.g., --from-uqff 8 finds q8_0-0.uqff or afq8-0.uqff). Also accepts ISQ type names (e.g., q4k, afq8). Shards are auto-discovered: specifying the first shard (e.g., q4k-0.uqff) automatically finds q4k-1.uqff, etc. Use semicolons to separate different quantizations | |
--isq-organization <ISQ_ORGANIZATION> | ISQ organization strategy: default or moqe | |
--imatrix <IMATRIX> | imatrix file for enhanced quantization | |
--calibration-file <CALIBRATION_FILE> | Calibration file for imatrix generation | |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
--paged-attn <MODE> | auto | PagedAttention mode - auto: enabled on CUDA, disabled on Metal/CPU (default) - on: force enable (fails if unsupported) - off: force disable Possible values: auto, on, off. |
--pa-context-len <CONTEXT_LEN> | Allocate KV cache for this context length. If not specified, defaults to using 90% of available VRAM | |
--pa-memory-mb <MEMORY_MB> | GPU memory to allocate in MBs (alternative to context-len) | |
--pa-memory-fraction <MEMORY_FRACTION> | GPU memory utilization fraction 0.0-1.0 (alternative to context-len/memory-mb) | |
--pa-block-size <BLOCK_SIZE> | Tokens per block (default: 32 on CUDA) | |
--pa-cache-type <CACHE_TYPE> | auto | KV cache quantization type |
--max-edge <MAX_EDGE> | Maximum edge length for image resizing (aspect ratio preserved) | |
--max-num-images <MAX_NUM_IMAGES> | Maximum number of images per request | |
--max-image-length <MAX_IMAGE_LENGTH> | Maximum image dimension for device mapping | |
--no-kv-cache | false | Disable KV cache entirely |
--matformer-config-path <MATFORMER_CONFIG_PATH> | Path to a MatFormer config (CSV/JSON describing available slices). See model card | |
--matformer-slice-name <MATFORMER_SLICE_NAME> | MatFormer slice to load (must match a slice name in the config file) | |
--mtp-model <MTP_MODEL> | MTP assistant model id or path | |
--mtp-n-predict <MTP_N_PREDICT> | Number of MTP draft tokens to propose per target step | |
--adapter <ADAPTER> | LoRA adapter alias to benchmark. Omit to benchmark the base model | |
--prompt-len <PROMPT_LEN> | 512 | Input lengths used to measure time to first token. Zero skips TTFT. Accepts comma-separated values for sweeps |
--gen-len <GEN_LEN> | 128 | Output tokens per decode request. Values below 2 skip decode metrics |
--depth <DEPTH> | 4 | Input context lengths used to measure decode TPOT. Accepts comma-separated values for sweeps |
--iterations <ITERATIONS> | 3 | Number of benchmark iterations |
--warmup <WARMUP> | 1 | Number of warmup runs per benchmark case (discarded) |
mistralrs bench auto
Section titled “mistralrs bench auto”Auto-detect model type (recommended)
mistralrs bench auto [OPTIONS] --model-id <MODEL_ID>| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | required | HuggingFace model ID or local path to model directory |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--format <FORMAT> | Model format: plain (safetensors), gguf, or ggml Auto-detected if not specified Possible values: plain, gguf, ggml. | |
-f, --quantized-file <QUANTIZED_FILE> | Quantized model filename(s) for GGUF/GGML (semicolon-separated for multiple) | |
--tok-model-id <TOK_MODEL_ID> | Model ID for tokenizer when using quantized format | |
--gqa <GQA> | 1 | GQA value for GGML models |
--enable-lora | false | Enable dynamic LoRA without preloading an adapter. Supports text models. Qwen3.5/3.6 MoE requires automatic model selection; vision-tower adapters are unsupported |
--lora <ALIAS=SOURCE|JSON> | Preload a language-model LoRA adapter as ALIAS=SOURCE. Remote adapters use revision main. May be repeated. Qwen3.5/3.6 MoE conditional-generation models require auto model selection; vision-tower adapters are unsupported | |
--lora-max-adapters <LORA_MAX_ADAPTERS> | 16 | Maximum loaded LoRA aliases and, independently, resident adapter generations |
--lora-max-rank <LORA_MAX_RANK> | 256 | Maximum rank accepted for a LoRA adapter |
--lora-max-bytes <BYTES> | 8589934592 | Maximum memory used by loaded adapters |
--legacy-lora <SOURCE> | Legacy LoRA adapter source for a raw GGUF or GGML model | |
--legacy-lora-order <LEGACY_LORA_ORDER> | Ordering JSON file for a legacy raw GGUF or GGML LoRA adapter | |
--xlora <XLORA> | X-LoRA adapter model ID | |
--xlora-order <XLORA_ORDER> | X-LoRA ordering JSON file | |
--tgt-non-granular-index <TGT_NON_GRANULAR_INDEX> | Target non-granular index for X-LoRA | |
--quant <QUANT> | Quantization front-door: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) This prefers prebuilt UQFF from mistralrs-community/<model>-UQFF, so use --isq if you do not want to switch to a prebuilt UQFF | |
--isq <IN_SITU_QUANT> | In-situ quantization: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) and quantizes the selected model in-place (in-situ) | |
--from-uqff <FROM_UQFF> | UQFF file(s) to load from. Accepts numeric shorthands (2, 3, 4, 5, 6, 8) to auto-detect the appropriate UQFF file (e.g., --from-uqff 8 finds q8_0-0.uqff or afq8-0.uqff). Also accepts ISQ type names (e.g., q4k, afq8). Shards are auto-discovered: specifying the first shard (e.g., q4k-0.uqff) automatically finds q4k-1.uqff, etc. Use semicolons to separate different quantizations | |
--isq-organization <ISQ_ORGANIZATION> | ISQ organization strategy: default or moqe | |
--imatrix <IMATRIX> | imatrix file for enhanced quantization | |
--calibration-file <CALIBRATION_FILE> | Calibration file for imatrix generation | |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
--paged-attn <MODE> | auto | PagedAttention mode - auto: enabled on CUDA, disabled on Metal/CPU (default) - on: force enable (fails if unsupported) - off: force disable Possible values: auto, on, off. |
--pa-context-len <CONTEXT_LEN> | Allocate KV cache for this context length. If not specified, defaults to using 90% of available VRAM | |
--pa-memory-mb <MEMORY_MB> | GPU memory to allocate in MBs (alternative to context-len) | |
--pa-memory-fraction <MEMORY_FRACTION> | GPU memory utilization fraction 0.0-1.0 (alternative to context-len/memory-mb) | |
--pa-block-size <BLOCK_SIZE> | Tokens per block (default: 32 on CUDA) | |
--pa-cache-type <CACHE_TYPE> | auto | KV cache quantization type |
--max-edge <MAX_EDGE> | Maximum edge length for image resizing (aspect ratio preserved) | |
--max-num-images <MAX_NUM_IMAGES> | Maximum number of images per request | |
--max-image-length <MAX_IMAGE_LENGTH> | Maximum image dimension for device mapping |
mistralrs bench text
Section titled “mistralrs bench text”Text generation model with explicit configuration
mistralrs bench text [OPTIONS] --model-id <MODEL_ID>| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | required | HuggingFace model ID or local path to model directory |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--format <FORMAT> | Model format: plain (safetensors), gguf, or ggml Auto-detected if not specified Possible values: plain, gguf, ggml. | |
-f, --quantized-file <QUANTIZED_FILE> | Quantized model filename(s) for GGUF/GGML (semicolon-separated for multiple) | |
--tok-model-id <TOK_MODEL_ID> | Model ID for tokenizer when using quantized format | |
--gqa <GQA> | 1 | GQA value for GGML models |
--enable-lora | false | Enable dynamic LoRA without preloading an adapter. Supports text models. Qwen3.5/3.6 MoE requires automatic model selection; vision-tower adapters are unsupported |
--lora <ALIAS=SOURCE|JSON> | Preload a language-model LoRA adapter as ALIAS=SOURCE. Remote adapters use revision main. May be repeated. Qwen3.5/3.6 MoE conditional-generation models require auto model selection; vision-tower adapters are unsupported | |
--lora-max-adapters <LORA_MAX_ADAPTERS> | 16 | Maximum loaded LoRA aliases and, independently, resident adapter generations |
--lora-max-rank <LORA_MAX_RANK> | 256 | Maximum rank accepted for a LoRA adapter |
--lora-max-bytes <BYTES> | 8589934592 | Maximum memory used by loaded adapters |
--legacy-lora <SOURCE> | Legacy LoRA adapter source for a raw GGUF or GGML model | |
--legacy-lora-order <LEGACY_LORA_ORDER> | Ordering JSON file for a legacy raw GGUF or GGML LoRA adapter | |
--xlora <XLORA> | X-LoRA adapter model ID | |
--xlora-order <XLORA_ORDER> | X-LoRA ordering JSON file | |
--tgt-non-granular-index <TGT_NON_GRANULAR_INDEX> | Target non-granular index for X-LoRA | |
--quant <QUANT> | Quantization front-door: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) This prefers prebuilt UQFF from mistralrs-community/<model>-UQFF, so use --isq if you do not want to switch to a prebuilt UQFF | |
--isq <IN_SITU_QUANT> | In-situ quantization: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) and quantizes the selected model in-place (in-situ) | |
--from-uqff <FROM_UQFF> | UQFF file(s) to load from. Accepts numeric shorthands (2, 3, 4, 5, 6, 8) to auto-detect the appropriate UQFF file (e.g., --from-uqff 8 finds q8_0-0.uqff or afq8-0.uqff). Also accepts ISQ type names (e.g., q4k, afq8). Shards are auto-discovered: specifying the first shard (e.g., q4k-0.uqff) automatically finds q4k-1.uqff, etc. Use semicolons to separate different quantizations | |
--isq-organization <ISQ_ORGANIZATION> | ISQ organization strategy: default or moqe | |
--imatrix <IMATRIX> | imatrix file for enhanced quantization | |
--calibration-file <CALIBRATION_FILE> | Calibration file for imatrix generation | |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
--paged-attn <MODE> | auto | PagedAttention mode - auto: enabled on CUDA, disabled on Metal/CPU (default) - on: force enable (fails if unsupported) - off: force disable Possible values: auto, on, off. |
--pa-context-len <CONTEXT_LEN> | Allocate KV cache for this context length. If not specified, defaults to using 90% of available VRAM | |
--pa-memory-mb <MEMORY_MB> | GPU memory to allocate in MBs (alternative to context-len) | |
--pa-memory-fraction <MEMORY_FRACTION> | GPU memory utilization fraction 0.0-1.0 (alternative to context-len/memory-mb) | |
--pa-block-size <BLOCK_SIZE> | Tokens per block (default: 32 on CUDA) | |
--pa-cache-type <CACHE_TYPE> | auto | KV cache quantization type |
mistralrs bench multimodal
Section titled “mistralrs bench multimodal”Multimodal model
mistralrs bench multimodal [OPTIONS] --model-id <MODEL_ID>| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | required | HuggingFace model ID or local path to model directory |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--format <FORMAT> | Model format: plain (safetensors), gguf, or ggml Auto-detected if not specified Possible values: plain, gguf, ggml. | |
-f, --quantized-file <QUANTIZED_FILE> | Quantized model filename(s) for GGUF/GGML (semicolon-separated for multiple) | |
--tok-model-id <TOK_MODEL_ID> | Model ID for tokenizer when using quantized format | |
--gqa <GQA> | 1 | GQA value for GGML models |
--quant <QUANT> | Quantization front-door: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) This prefers prebuilt UQFF from mistralrs-community/<model>-UQFF, so use --isq if you do not want to switch to a prebuilt UQFF | |
--isq <IN_SITU_QUANT> | In-situ quantization: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) and quantizes the selected model in-place (in-situ) | |
--from-uqff <FROM_UQFF> | UQFF file(s) to load from. Accepts numeric shorthands (2, 3, 4, 5, 6, 8) to auto-detect the appropriate UQFF file (e.g., --from-uqff 8 finds q8_0-0.uqff or afq8-0.uqff). Also accepts ISQ type names (e.g., q4k, afq8). Shards are auto-discovered: specifying the first shard (e.g., q4k-0.uqff) automatically finds q4k-1.uqff, etc. Use semicolons to separate different quantizations | |
--isq-organization <ISQ_ORGANIZATION> | ISQ organization strategy: default or moqe | |
--imatrix <IMATRIX> | imatrix file for enhanced quantization | |
--calibration-file <CALIBRATION_FILE> | Calibration file for imatrix generation | |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
--paged-attn <MODE> | auto | PagedAttention mode - auto: enabled on CUDA, disabled on Metal/CPU (default) - on: force enable (fails if unsupported) - off: force disable Possible values: auto, on, off. |
--pa-context-len <CONTEXT_LEN> | Allocate KV cache for this context length. If not specified, defaults to using 90% of available VRAM | |
--pa-memory-mb <MEMORY_MB> | GPU memory to allocate in MBs (alternative to context-len) | |
--pa-memory-fraction <MEMORY_FRACTION> | GPU memory utilization fraction 0.0-1.0 (alternative to context-len/memory-mb) | |
--pa-block-size <BLOCK_SIZE> | Tokens per block (default: 32 on CUDA) | |
--pa-cache-type <CACHE_TYPE> | auto | KV cache quantization type |
--max-edge <MAX_EDGE> | Maximum edge length for image resizing (aspect ratio preserved) | |
--max-num-images <MAX_NUM_IMAGES> | Maximum number of images per request | |
--max-image-length <MAX_IMAGE_LENGTH> | Maximum image dimension for device mapping |
mistralrs bench diffusion
Section titled “mistralrs bench diffusion”Image generation model (diffusion)
mistralrs bench diffusion [OPTIONS] --model-id <MODEL_ID>| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | required | HuggingFace model ID or local path to model directory |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
mistralrs bench speech
Section titled “mistralrs bench speech”Speech synthesis model
mistralrs bench speech [OPTIONS] --model-id <MODEL_ID>| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | required | HuggingFace model ID or local path to model directory |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
mistralrs bench embedding
Section titled “mistralrs bench embedding”Embedding model
mistralrs bench embedding [OPTIONS] --model-id <MODEL_ID>| Option | Default | Description |
|---|---|---|
-m, --model-id <MODEL_ID> | required | HuggingFace model ID or local path to model directory |
-t, --tokenizer <TOKENIZER> | Path to local tokenizer.json file | |
-a, --arch <ARCH> | Model architecture (auto-detected if not specified) | |
--dtype <DTYPE> | auto | Model data type |
--format <FORMAT> | Model format: plain (safetensors), gguf, or ggml Auto-detected if not specified Possible values: plain, gguf, ggml. | |
-f, --quantized-file <QUANTIZED_FILE> | Quantized model filename(s) for GGUF/GGML (semicolon-separated for multiple) | |
--tok-model-id <TOK_MODEL_ID> | Model ID for tokenizer when using quantized format | |
--gqa <GQA> | 1 | GQA value for GGML models |
--quant <QUANT> | Quantization front-door: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) This prefers prebuilt UQFF from mistralrs-community/<model>-UQFF, so use --isq if you do not want to switch to a prebuilt UQFF | |
--isq <IN_SITU_QUANT> | In-situ quantization: accepts numeric levels (2, 3, 4, 5, 6, 8) or raw quant names (q4k, q8_0, etc.) and quantizes the selected model in-place (in-situ) | |
--from-uqff <FROM_UQFF> | UQFF file(s) to load from. Accepts numeric shorthands (2, 3, 4, 5, 6, 8) to auto-detect the appropriate UQFF file (e.g., --from-uqff 8 finds q8_0-0.uqff or afq8-0.uqff). Also accepts ISQ type names (e.g., q4k, afq8). Shards are auto-discovered: specifying the first shard (e.g., q4k-0.uqff) automatically finds q4k-1.uqff, etc. Use semicolons to separate different quantizations | |
--isq-organization <ISQ_ORGANIZATION> | ISQ organization strategy: default or moqe | |
--imatrix <IMATRIX> | imatrix file for enhanced quantization | |
--calibration-file <CALIBRATION_FILE> | Calibration file for imatrix generation | |
--cpu | false | Force CPU-only execution |
-n, --device-layers <DEVICE_LAYERS> | Device layer mapping (format: ORD:NUM;… e.g., “0:10;1:20”) Omit for automatic device mapping | |
--topology <TOPOLOGY> | Topology YAML file for device mapping | |
--hf-cache <HF_CACHE> | Custom HuggingFace cache directory | |
--max-seq-len <MAX_SEQ_LEN> | 4096 | Max sequence length for automatic device mapping |
--max-batch-size <MAX_BATCH_SIZE> | 1 | Max batch size for automatic device mapping |
--paged-attn <MODE> | auto | PagedAttention mode - auto: enabled on CUDA, disabled on Metal/CPU (default) - on: force enable (fails if unsupported) - off: force disable Possible values: auto, on, off. |
--pa-context-len <CONTEXT_LEN> | Allocate KV cache for this context length. If not specified, defaults to using 90% of available VRAM | |
--pa-memory-mb <MEMORY_MB> | GPU memory to allocate in MBs (alternative to context-len) | |
--pa-memory-fraction <MEMORY_FRACTION> | GPU memory utilization fraction 0.0-1.0 (alternative to context-len/memory-mb) | |
--pa-block-size <BLOCK_SIZE> | Tokens per block (default: 32 on CUDA) | |
--pa-cache-type <CACHE_TYPE> | auto | KV cache quantization type |